ABSTRACT
This article addresses why the literature on Inequality of Educational Opportunity (IEO) reaches diverging results concerning the decline or persistency of IEO over time. The main argument in this article is that the diverging results may be caused by the fact that the social class variables used to capture trends in IEO act as proxies for unobserved family-background influences that are substantively different from social class. The article analyses extremely rich longitudinal data from Denmark spanning three generations within the same family lineage. It demonstrates, first, that the effect of social class on secondary schooling is overstated when other family influences, conceptualised as economic, cultural and social capital, and unobserved family influences are not taken into consideration, and second, as in the other Scandinavian countries, that IEO has declined significantly in the postwar period.
Acknowledgements
The author would like to thank two anonymous reviewers from European Societies, Anders Holm, Martin Munk, Signe Hald Andersen and Jens Bonke for generous comments. This research was sponsored by the Danish National Research Council and the Danish National Institute of Social Research.
Notes
1The literature on social mobility is comprehensive. For reviews, see, e.g., Treiman and Ganzeboom (Citation1990); Ganzeboom et al. (Citation1991); Erikson and Goldthorpe (Citation2002); Solon (Citation2002); Breen (Citation2004); Breen and Jonsson (Citation2005).
2Mothers in the DYLS analysis sample were on average 22.8 years old at the birth of their first child. Fathers were on average 24.1 years old. In the total DYLS sample, mean age at birth of first child is 25.1 for mothers and 28.0 for fathers, while national mean age at first birth for women in the period 1975–1985 was around 24.7 (Statistics Denmark).
3The mixed logit model might also be written separately for each of the two generations in which case the g subscript is omitted from Equation (1). However, I use the compact form to conserve space and to underscore, first, that the model is estimated simultaneously for both generations and, second, that the random effects capturing unobserved traits in each generation are allowed to be correlated or ‘shared’ across generations.
4If the model is run without parental income, home ownership, car ownership and ownership of a summer house become significant at around p=0.05.